Engine body vibration is an important safety concern for diesel engines as it can lead to severe engine damages and even catastrophic failures. Traditional methods for excitation computation assume that the rotational speed and the cylinder pressure remain stable under each working cycle, which makes it impossible to precisely predict vibration peaks at certain frequencies. But in practice, the existence of shafting torsional vibration will lead to the fluctuation of rotational speed and cylinder pressure. To predict engine block vibration more accurately, this paper proposes a novel method considering the instantaneous variations in rotational speed and cylinder pressure. It is implemented by two steps. Firstly, an innovative model considering the coupling effect between shaft torsional vibration and advanced injection angle is proposed, it particularly explains how the cylinder pressure and instantaneous rotational speed vary due to the coupling effects. Secondly, the excitation forces that cause block vibration are analyzed and their computational formulas are improved. This method also considers the offset of excitation loading position caused by torsional vibration. The coupling model is experimentally verified and the excitations are loaded into a finite element model of the diesel engine for vibration response prediction. The results reveal that the coupled model can predict vibration peaks at sideband frequencies and thus makes the frequency spectrum much richer than that of the traditional method. Furthermore, it shows that the acceleration results obtained by the coupled model have better prediction accuracy than those obtained from traditional methods, with the deviation percentage reducing from 6.31% and 8.41% to 0.64% and 0.32% in the X and Z direction. It indicates that the coupled model is more reliable in predicting the vibration response of the engine block. Finally, the effect of advanced injection angle on engine block vibration is investigated and the optimal value of the advanced injection angle is determined to be 29 CA, which makes the engine block vibration reduced by 3.60%, 2.72%, and 3.16% in the X, Y, and Z direction comparing to the original advanced injection angle. Overall, this study presents an improved approach for predicting engine block vibrations and offers valuable insights for the design of shafting structures and fuel injection parameters in marine diesel engines.